Docs / The Opportunity Finder

The Opportunity Finder

Most buying mistakes are one of two: chasing a stock at its highs, or catching a falling knife because it looks cheap. The Finder is built to make both hard to do.

Its rule fits in a sentence: qualify on strength, buy on weakness. Decide which companies deserve to be bought while they're strong — then wait, sometimes for weeks, until one of them dips. Never the other way round.

The universe — ~100 quality namesQualified — the watchlistEvaluated — why did it drop?Selectedonly the strong passonly if it really dippedonly a healthy dip passestrend · quality · strength vs market · momentuma real fall from its recent highno company damage · at most three a day
Each arrow is a gate; most names stop at one of them. What reaches the bottom earns the approval email — and your verdict.

Gate one: Qualified — strong enough to deserve a dip

Every day the Finder scores a universe of about a hundred large companies on four lenses: trend (above its long-term lines), business quality, strength relative to the market over recent months, and momentum. Only names that clear the composite bar become Qualified — a watchlist of stocks the system would want if they got cheaper. Weakness alone never gets a stock onto this list.

Exactly what happens at gate one — step by step
  1. At 16:20 the universe refreshes: the Nasdaq-100 list, the day's prices, and fundamentals for every name.
  2. Each stock is scored out of 100 on the four lenses (July 2026 weights: trend 35, business quality 25, relative strength 20, momentum 20). Trend means the moving averages stacked in order — 20-day above 50-day above 200-day above 50-week — with the price above the 200-day line (revised August 2026); quality means profitability and balance-sheet checks — and a name with incomplete data is excluded, never guessed; relative strength means beating the S&P 500 over the last 63 trading days; momentum means RSI above 50 with no short-term breakdown.
  3. Score 80 or better → Qualified. The watchlist is capped at 15 names, so the bar rises with the competition.
  4. Guards apply even to the strong: a stock stretched too far above its 200-day line, up parabolically over three months, or with RSI at overbought extremes is qualified but not buyable — strength, yes; froth, no.
  5. Strength must be maintained, not achieved once: a qualified name whose averages un-stack, or that closes below its 200-day line, is disqualified again. A dip through the 50-day line alone is tolerated — often that is the buyable dip (it must only be above the 50-day on the day it first qualifies).
  6. Then — usually — nothing. The name sits in waiting for a pullback, sometimes for weeks. That is the gate's normal, healthy state.

Gate two: Evaluated — it dipped; now, why?

When a qualified name falls a real distance from its recent high, a buy signal forms — and immediately meets the question depth can't answer: why did it drop? The evaluator reads the day's headlines and the market around it. A stock down because the whole market or its sector is down is cheaper. A stock down on its own bad news is damaged — and damage is rejected outright, however deep the discount. "Do not select a stock only because it dropped" is written into the strategy file itself.

Exactly what happens at gate two — step by step
  1. The trigger: a qualified name closes a real distance below its 20-day high (July 2026: at least 2%). A buy signal forms — but nothing is offered yet.
  2. Shortly after the engine run, the evaluator takes all of the day's signals as one batch. Every candidate is examined; none skips the queue on depth alone.
  3. Context first: if the whole market fell hard that day (S&P 500 down 1.5%+) or the stock's sector did (2%+), the drop is context — the good kind of cheap. Neither? Then the suspicion is company-specific.
  4. Then the news: the day's headlines for the name are read and classified. Guidance cuts, accounting trouble, lawsuits — real damage is rejected outright, whatever the discount.
  5. Even panic is measured: a healthy oversold reading is acceptable; a collapse-grade one is treated with suspicion, not excitement.
  6. Every survivor gets a dip-quality score, 0–100, and a one-sentence reason — both written to the ledger. You can read that exact sentence later in the trace, under Why Evaluated.
  7. Verdicts don't linger: yesterday's yes never outvotes today's no. A name demoted since its last evaluation is out.

Gate three: Selected — worth your attention today

What survives is scored for dip quality, and only the highest scores — at most three a day — reach the approval email. Deeper dips get more capital, inside hard caps: per purchase, per name, and per day, so that even a generous day of signals can't overcommit the book. The rest wait; if their dip holds tomorrow, they're scored again.

Exactly what happens at gate three — step by step
  1. Selection: dip-quality score of 70 or better (July 2026), and only the top three of the day. Fourth-best waits for tomorrow.
  2. Two vetoes can still strike: an imminent earnings dateblocks the buy (however good the dip — no coin-flips on announcement night), and a broken market regime blocks everything.
  3. Sizing follows depth: a 2–4% dip commits 5% of the agent account's available cash, 4–6% commits 7%, deeper commits 10% — inside hard caps per purchase, per name, and at most ~30% of cash across the day's buys combined.
  4. 16:38 — the Opportunity Brief tells the whole funnel story: Selected, Evaluated, Qualified, each name with its reason.
  5. 16:40 — the Approval email carries the Selected with their sizes. This is the only actionable moment; everything before it was information.
  6. You approve → the order executes on the agent book, the fill is captured, a ledger row opens — graded daily from that moment — and the 20-trading-day time-stop starts counting.

Where it buys, and when it lets go

Finder buys execute only on the agent account — the experiment runs on the experimental book. The market regime gates everything: when the tape is broken, no new buys at all. And every dip-buy carries the 20-trading-day time-stop from the Control Tower: a dip thesis that hasn't worked in a month isn't working.

The exact thresholds — dip depth, scores, sizing tiers, caps — live on the strategy card on Home, straight from the rules file in force. What's written here is the shape; what's written there is today's truth.

The Finder's discipline is refusal. On most days its best work is the list of things it did not offer you.

Read on

Both engines end at the same place: an ask, and your verdict. The glossary → defines every word they use on the way there.